PRODUCT LAUNCH DECISIONS BY DOMINANT AND FRINGE FIRMS

John M. de Figueiredo* Margaret K. Kyle** MIT Carnegie Mellon

26 February 2004

Abstract

In this paper we analyze the product launch decisions by dominant and fringe firms. In particular, we explore three questions in the product entry process: which firms enter, when they enter, and what happens after they enter. To do this, we use a database of all product launches in the desktop laser printer industry from 1984 to 1997. We show that dominant firms are quite selective in the submarkets they pursue. In particular, they often stay out of the technologically advanced markets, and choose to enter markets well behind the technological frontier set by fringe firms. When dominant firms do enter submarkets, they are usually early entrants, launching products before many fringe firms. Additionally, we analyze what happens to markets after entry by dominant firms. Many antitrust scholars and policy-makers fear that dominant firm entry into technology markets forces out fringe competitors, leading to high market power for the dominant entrant. We find that both entry and exit rates increase after a dominant firms enters a submarket, but that there is a net increase on average in the number of competing models and substantial growth in demand. In addition, while dominant firms introduce models priced at a premium relative to the average in a product class, the average price within that class tends to decline more as a result of competition from dominant firms than from fringe firms.

* Sloan School of Management, Massachusetts Institute of Technology E52-546, 50 Memorial Drive, Cambridge, MA 02142-1347, 617-258-7253, jdefig@mit.edu ** Corresponding Author: Graduate School of Industrial Administration, Carnegie Mellon University, 5000 Forbes Ave, Pittsburgh, PA 15213, 412-268-6820, mkkyle@cmu.edu

1

PRODUCT LAUNCH DECISIONS BY DOMINANT AND FRINGE FIRMS

John M. de Figueiredo and Margaret K. Kyle 26 February 2004

I. INTRODUCTION

The evolution of an industry is determined in part by the product launch decisions

firms make. Firms must make two key decisions regarding new product launch: whether

to enter a market with a new product and when to launch that product. On both of these

questions, the theory and evidence are mixed on what drives these decisions. An

extensive theoretical literature on the relationship between market structure and

innovation contends that incumbent firms often face different incentives for innovation

than do de novo entrants. Depending on the nature of the innovation (in particular,

whether it is radical or incremental, or how easily imitated), incumbent firms are

sometimes expected to be innovators – or early entrants into a new technical subfield or

product market -- and other times to be followers. Likewise, previous empirical work on

these questions has mixed findings. In some industries, incumbents are shown to be at

the leading edge of innovation, at least as measured by R&D intensity and patent output

(Mansfield 1964, Grabowski 1968, Scherer 1965). In others, they are shown to be second

or late movers, such as in Christensen (1994), Lerner (1997) and Henderson (1989).

One source of these mixed results is that much of the empirical work on

technology races or the timing of incumbent entry has not separately considered the

firm’s decision to enter a new submarket (or, alternatively, to participate in a

technological race) from its decision when to enter. That is, the timing of market entry is

usually considered without conditioning on entry. However, many firms that could

2

potentially enter the market, choose not to do so, so an examination of the timing of

market entry must also consider the selection process of those entrants.1

In this paper, we consider the decision to enter and the timing of entry conditional

on the decision to enter for nearby or submarket niches. We address three questions.

First, which submarkets do dominant firms avoid, but fringe firms enter? Second, are

dominant firms the first movers in the submarkets they choose to enter? Finally, we look

at the effects of entry by a dominant firm into a submarket it has chosen to enter. Does

the presence of a dominant firm reduce the rate of subsequent entry or increase the

likelihood of a fringe firm’s failure in a submarket? This last question relates not only

to market structure, but also has important implications for the strategy of these fringe

firms. Indeed, in answering this last question, we consider the full cycle of the entry

process: which firms enter, when they enter, and what happens after they enter. Thus,

we are able to use fine-grained measures of multi-dimensional performance to explore the

consequences of entry by dominant firms on prices and market structure.

We examine these predictions in the context of the desktop laser printer industry

by comparing the behavior of dominant and fringe firms for 14 years after the inception

of the industry. The desktop laser printer industry is particularly appropriate for

understanding these questions. First, the performance of laser printers has consistently

improved over time and can be easily observed. Second, we are able to track the entry

and exit of nearly every product in the industry since its inception in 1984 through 1996.

Third, the competitive environment varies across the product space and over time,

providing some identification power. Fourth, there are heterogeneous firms of different

sizes and with varied backgrounds. Finally, and perhaps most importantly, the laser 1 Mitchell (1989) considers such a model, which is discussed in the theory section.

3

printer industry shares many characteristics with other high technology industries, such as

personal and mainframe computers, disk drives, fax machines, retail electronics, digital

cameras, and the like. The products are differentiated; there is an innovation frontier;

there is an important mass market; and product and firm turnover is prevalent. Like other

industries, technical advances frequently give rise to market opportunities. These factors

affect firms of all sizes, including both incumbents and entrants. Thus the insights from

this study may be applicable to broad sectors of the economy.

In this paper, we demonstrate how entry is affected by, and affects, market

structure and the firm’s antecedent market position. We show that dominant firms are

quite selective in the submarkets they pursue. In particular, they often stay out of the

technologically advanced markets, and choose to enter markets well behind the

technological frontier set by fringe firms. When dominant firms do enter submarkets,

they are usually early entrants, launching products before many fringe firms.

Additionally, we analyze what happens to markets after entry by dominant firms. Many

antitrust scholars and policy-makers fear that dominant firm entry into technology

markets forces out fringe competitors, leading to high market power for the dominant

entrant. We find that both entry and exit rates increase after a dominant firm enters a

submarket, but that there is a net increase on average in the number of competing models

and substantial growth in demand. In addition, while dominant firms introduce models

priced at a premium relative to the average in a product class, the average price within

that class tends to decline more as a result of competition from dominant firms than from

fringe firms.

4

In the next section, we outline the theoretical and empirical debate that exists, and

the omissions that occur. In Section III, we outline our data, method, and approach.

Section IV presents our results. We conclude in Section V.

II. THEORY: TWO DEBATES AND TWO OMISSIONS

Although the literature on entry and innovation is extensive, there are two central debates

in the literature: the entry decision and the timing of entry.

A. THE ENTRY DECISION

The first debate centers on the entry decision, and whether dominant or incumbent

firms choose to enter new innovative markets. The theory and evidence from the

economics, marketing, and innovation literature is mixed. One view of this question in

the economics and innovation literature is that of Reinganum (1989), who considers a

model in which a new innovation completely displaces the existing product. In this case,

the incumbent has less incentive than an entrant to invest in R&D because doing so

hastens the end of its existing market. Entrants are therefore likely to displace the

incumbent in markets with technology races of this type. In contrast, Gilbert and

Newbery (1982) treat innovations as incremental improvements to existing products.

Under their assumptions, no technology race occurs at all, and the incumbent is never

displaced. Empirical work finds support for both approaches, depending upon the

market. For example, Lerner (1997) finds that disk drive makers, who compete in a

“winner-take-all” market, behave in a way consistent with Reinganum’s model.

Henderson (1993) shows that incumbents successfully incorporate incremental

5

innovations into their products in photolithography equipment, but are overtaken by

newcomers to the industry when the innovation is radical. However, others find that

incumbents may be unwilling to extend their product line or brand because even

incremental innovations may dilute brand equity or result in a loss of share in core

markets (Randall et al 1998). Thus, even within the incremental innovation literature in

economics and marketing, predictions about the dominant firm’s incentive to innovate

and willingness to launch new products is mixed.

Similarly, the organizations literature provides a number of theories with different

implications for incumbent innovation. Several theories from this body of work suggest

that costs in the form of organizational disruption or reduced value of assets differ

between incumbents and entrants (among many others, Tushman and Anderson, 1986

and Hannan and Freeman, 1977). Essentially, fears of cannibalization, risk-aversion, and

organizational inertia all imply that the incumbent will not enter a new submarket, even if

incremental innovation is involved. However, if the incumbent perceives a threat to its

core products or owns specialized assets that effectively lower its entry costs into new

segments, it may be more likely to enter an emerging subfield (Brittain and Freeman,

1980). Clearly, variation in market conditions influences the outcomes, and as in the

economics literature, empirical studies in organizational behavior find cases of

incumbents quick to innovate as well as those that lag behind (Lambkin, 1988). Overall,

the decision of dominant firms to incrementally innovative to enter new markets, no

matter which literature, finds mixed, and at times conflicting, theory and evidence.

B. ORDER OF ENTRY

6

A second debate that continues in the entry literature occurs in understanding

the order of the entry—that is given a subset of firms have chosen to enter, are

incumbents first-movers or followers into new market segments. The literature has

characterized many factors that may influence the incentive of an incumbent firm to be an

early mover, rather than to delay. Technological leadership (learning curve, patents and

R&D), pre-emption of assets (preemption of input factors, preemption of geographic and

product characteristics space, preemptive plant and equipment), and buyer switching

costs (switching costs, buyer choice under uncertainty) are cited in different ways

throughout the literature as reasons for incumbent firms to be first-movers (Lieberman

and Montgomery 1988). For example, Nelson and Winter (1982) and Kamien and

Schwartz (1972) argue that incumbents enter submarkets more quickly when they face

potential competition. This is related to Lieberman’s (1987) argument that opportunities

for strategic preemption – through customer lock-in or network externalities, for example

– make incumbents early movers, and that the threat of being preempted has the same

effect (Wernerfelt and Karnini, 1987).

On the other hand, delay may be a dominant strategy under many conditions.

The ability to “free-ride” on first-mover investments, resolution of technological and

market uncertainty, and various types of “incumbent inertia” that make it difficult for the

incumbent to adapt to environmental change all may lead a firm to delay entry into a new

subfield (Lieberman and Montgomery 1988). For example, if a product can be easily

imitated, Katz and Shapiro (1987) argue that a dominant firm is better off by being a

follower into a new submarket. The delay permits the dominant firm to imitate the first

mover’s product and avoids having its own products imitated. In addition, incorporating

7

innovations can be risky for firms that have a successful formula for their existing

products. Singh (1986) and March (1988) assert that in such a situation, incumbents may

delay if managers are risk-averse. Note that entrants, which do not face the same

organizational constraints, have no such disincentive. In fact, if firms take on more risk

when facing bankruptcy or failure (Bowman, 1982), fringe firms that are “just hanging

on” with a slim share are more likely to make the first high-risk move into a new market

segment.

Empirical work on order of entry provides support for both sides of the argument.

(see Robinson 1988, Robinson et al 1992, or Kalyanaram et al 1995 for a review of the

literature) Namely, a number of studies show dominant or incumbent firms are likely to

enter first (Grabowski 1968, Schmalensee 1978, Bond and Lean 1977, Moore et al 1991),

while many other studies show dominant or incumbent firms are likely to hold back and

be later entrants (e.g. Henderson 1989, Lerner 1997, Glazer 1985, Mansfield et al 1981).

However, all of these empirical studies examine the order of entry conditional on the

decision to enter, without accounting for the selection effect of that choice. Lieberman

and Montgomery (1988) and Goldner and Tellis (1993), in their studies of first-mover

advantages, noted that this empirical approach leads to a sample selection bias problem,

because only those who enter are observed. In our paper, we estimate the order of entry

decision as a two-stage process—the decision to enter and then the order of entry, to

correct for this empirical problem. Only Mitchell (1989) has used this approach in

estimating entry models. This paper examines incumbent behavior in the medical

imaging industry over several decades and finds that conditional on their decision to

enter, incumbent firms were early movers into new submarkets. However, incumbent

8

firms often elected to stay out of new markets, particularly if their core products were not

threatened or it they lacked a competitive advantage over new firms in such a market.

C. OMISSIONS

In addition to the sample selection problem just discussed, there are a number of

alternative explanations for entry behavior that have largely been omitted from empirical

analysis, but which may be quite important. Potential demand is unobserved by the

researcher, and firms have incomplete information. If incumbency leads to better

knowledge of market conditions and identification of the best submarkets to enter, the

same entry pattern would result. (Adner and Levinthal 2002). Similarly, competition

may play out as a Stackelberg game in which the dominant firms move first, choosing the

segment that is most profitable to them, and the fringe firms follow and play

differentiated Bertrand (i.e. differentiate in product space). This would also give the

same pattern. Previous papers have not considered these alternative explanations; we

hope to begin to make headway in addressing these issues.

This latter point leads to a second omission in the empirical literature. After entry

by dominant firms, whether early or late, what happens to an industry? The primary

argument in the strategic management, innovation management, and marketing literature

is that holding everything else constant, once a dominant firm enters, whether early or

late, other firms are more likely to stay out of the market niche, and those already in the

market niche are more likely to exit. This is because the dominant firm exerts its market

power through a superior brand or lower cost, and fringe firms fail as a result (Bowman

9

and Gatignon 1996) even when there is easy imitation by followers ro those already in

the industry (Makadok 1998).2

However, there is an alternative rationale that has not been explored. Entry by a

dominant firm actually signals something about the market conditions that fringe firms

cannot observe. For example, it might be (as above) that dominant firms have better

information about market conditions, demand, and potential profitability of a market

segment. Thus, dominant firm entry does not drive fringe firms out of markets, but

actually attracts them into the market. Far from having a detrimental competitive effect,

dominant firm entry may actually have a positive competitive effect in the market, ceteris

paribus.

Our paper seeks to contribute to all three areas by addressing the entry decision,

the timing of entry, and the effects of entry by dominant firms on competition, using data

from the laser printer industry.

III. DATA AND METHOD

A. THE DESKTOP LASER PRINTER INDUSTRY

As the personal computer market expanded in 1980s, so too did the market for

desktop printers. Hewlett-Packard introduced the first desktop laser printer for the retail

market in 1984. By the end of 1985, 17 firms had introduced 23 models of printers.

Figure 1 illustrates the number of firms and models in the industry from the beginning of

2 Mitchell (1991) argues that one must control not only for overall order of entry effect, but the incumbent order of entry into the subfield as well. This is because there are dual clocks that determine the competitive advantage of early movers.

10

the industry in 1984 to 1997. At its peak in 1990, the industry had 100 firms. Since that

time, the number of firms has fallen off to 87.

The information on laser printer characteristics, entry, and exit come from a

variety of sources. The primary source is Dataquest’s SpecCheck analysis of page

printers. Dataquest follows each manufacturer’s products and records a variety of

product characteristics, including ship date, speed, resolution, and other features. The

data were incomplete for many models, so we supplemented this data with information

from trade journals, private analysts’ reports, and general industry data provided to us by

a private consulting firm. We believe the dataset, which covers the industry from its

inception in 1984 to 1997, is the most comprehensive available. Over this 14-year

period, we are able to record 2,928 printer-year observations. Though we have attempted

to be as thorough as possible, there remain some printers for which we cannot identify all

of the independent variables. These have been dropped from the analysis.

Three types of firms populate the industry. Ricoh, IBM, Hewlett Packard, Canon,

and Xerox are examples of large, diversified firms with a strong presence. A number of

medium-sized firms specialize in multiple printer technologies, such as Lexmark,

Kyocera, Genicom, and Kentek. Finally, there are over 100 very small “fringe” firms,

which produce few printer models, ship very few units, and tend to appear in the industry

only briefly. Hewlett Packard is the dominant firm in the industry, and has maintained

between 45% and 65% market share for most of the industry’s history. Defining a

dominant firm as one that has greater than 10% market share, we find that dominant firms

account for between 51% and 83% of the total market share, but they account for only

1% and 12% of the number of products introduced.

11

We treat each printer with a unique model number as a distinct product whose

features do not change after introduction.3 The number of products on the market has

generally been increasing over time, as seen in Figure 2, with a peak at 855 product

models in 1996, while the number of firms has begun to decline. The average number of

products per firm was 8.8 in 1996, up from 1.8 in 1988.4

B. VARIABLE DEFINITIONS

The variable definitions and descriptive statistics are found in Table 1. The

average product stays on the market for four years, and costs almost $3,983. The average

speed of printers is 10 PPM, and the average resolution is 450 dpi. Although printers can

be characterized on a number of dimensions, the two most common measures of printer

performance are speed, measured by pages per minute (PPM), and resolution, measured

by dots per inch (DPI). Printers are bunched tightly in groups in the performance space.

Figure 3 shows the distribution of printer models across the 20 discrete product classes

(or niches) in terms of these two characteristics that we defined based on the clear

groupings of printers. We provide some data on these classes in Table 1.

C. METHOD

In addition to the graphical evidence and descriptive statistics used above to

illustrate patterns in the industry, we employ some econometric specifications for more 3 Unlike some product markets, firms in this industry do not change printer attributes once the product has been introduced. Rather, they introduce new products. 4 In defining the industry, we appealed to the data and to industry experts and trade journals. These sources consistently define the desktop laser printer industry as laser printers that print 0-20 PPM, can be attached to a personal computer, and are small enough to fit on a desk. This industry definition has remained constant over the time period. Our statistical analyses are robust to small definitional changes.

12

refined statistics tests. Most of the methods used are standard in both the economics and

the organizations literature, namely linear regressions and Poisson regressions for count

variables. We also employ a bivariate probit analysis with sample selection in order to

examine questions related to the joint decision to enter or not enter, and the subsequent

timing of entry conditional on entering. Because this method is not common in the

literature, we describe it a little more depth here. We specify the following structural

model. Let the decision to enter or not enter (entry equation or selection equation) be

modeled as a first stage probit of the following form:

I Zi i i* = +γ ε

such that I II

i i

i

= >=

1 00 if otherwise

*

and whether you choose to enter early or late (timing equation) be modeled as a second

probit equation of the form:

Y X uj j j* = +β

such that Y YYj i

j j

j

= >

=

∈

1 00 if otherwise

*

We define, Ii* as the latent variable that measures the perceived probability of the firm

profiting from entry. Above some threshold, we observe entry, as indicated by Ii = 1 .

Otherwise we observe no entry, Ii = 0 . Whether a firm enters or not is determined by

13

the explanatory variables in the matrix Zi (discussed below). In the timing equation, Yj*

is the relatively profitability of early entry. When profitability of early entry is

sufficiently high, conditional on entry, the firm enters early. This is given by the observed

Yj = 1. If the firm perceives low profitability or losses to entering early, then we

observe Yj = 0 , late entry. The probability of entering early is determined by the

exogenous variables in matrix X j (described below), where Z Xi j≠ . We assume that

the error terms, εi and u j are jointly normally distributed with mean (vector) zero and a

variance-covariance matrix of the form:

∑ =⎡

⎣⎢

⎤

⎦⎥

11ρ

ρ

so that E ui j( , )ε ρ= . Thus there may be sample selection bias. The timing of entry is

correlated with your decision to enter. This two equation probit model, then

characterizes the structural model for the enter v. don’t enter and enter early v. enter late

equations.

We can estimate this model using a bivariate probit with adjustment for sample

selection bias (Van de Ven 1981). Assuming that εi and u j are bivariate standard

normally distributed with correlation coefficient ρ , the univariate cumulative distribution

function isΦ , and the bivariate cumulative normal distribution function is Φ2 , the

likelihood function to be maximized is:

( ) ( ) ( )L X Z X Z ZI Y

j iI Y

j iI

i= • − •= = = = =∏ ∏ ∏Φ Φ Φ2

1 12

1 0 0, ,

, ; , ; ,β γ ρ β γ ρ γ

14

This accounts for the three possible outcomes. The first term encompasses

observations where there has been entry and entry is early. The second term

encompasses those observations where there is entry and entry is late by the firm. The

final term includes those observations where there is no entry.5 These three terms then

pick up all the possible outcomes. This estimation technique allows us to jointly estimate

a two equation probit model, adjusting for sample selection bias.

IV. EMPIRICAL RESULTS

Table 2 contains three panels to address the question of which submarkets

dominant firms choose to compete in. The first panel shows the average of the “best”

products made by several firms which achieved at least 10% market share in one year.

The top of this panel presents the means of the maximum DPI and PPM of each firm’s

product portfolio during the years in which it was dominant, and the bottom of the panel

presents the same means calculated over all years in which the firm was active. It is clear

that the best product of the set of fringe firms has significantly higher DPI, but about the

same speed. Figures 4 and 5 demonstrate the same point graphically: it is evident that

some (though not all) fringe products are at the top of the DPI frontier, which the

products of dominant firms are clustered at lower levels of resolution. However, the

distribution of PPM is similar for both dominant and fringe firms.

The middle panel of Table 2 shows results from a regression of a firm’s best DPI

or PPM in each year on a few firm and market characteristics and year fixed effects.

With such a small sample size, the results should be interpreted with some caution;

5 We can say that: Prob (Y = 1/ I = 1) + Prob (Y = 0/ I = 1) + Prob (I = 0) = 1

15

however, it appears that dominant firms are significantly behind the best fringe firms on

both product characteristics.

Finally, the bottom panel of Table 2 shows the number of years firms were able to

stay at the frontier, or lead the industry in terms of technology. There appears to be quite

a bit of turnover at the top for resolution, a pattern similar to that in disk drives

established by Lerner. However, the top players in speed are much more stable. We can

only speculate at this point why we see this divergence. Perhaps firms did not engage in

racing behavior on the speed dimension because during this period, laser printing

technology was significantly slower than dot-matrix technology. Consumers who valued

speed would likely not be in the laser printer market at all.

Table 3 presents results from the bivariate probit model with sample selection.

These specifications address our second question of whether dominant firms are first-

movers. The first stage estimates the probability of entry into a new submarket, and the

second stage estimates the probability that a firm is the first mover into that submarket,

conditional on entry. Results from the first stage indicate that, consistent with Mitchell’s

findings, a firm that has a specialized asset that provides a competitive advantage in other

market segments (i.e., has received an editorial award for one of its products) is more

likely to launch a product in a new niche. Dominant firms are less likely to introduce a

product in a new submarket, all else equal. Conditional on entry, however, they tend to

be the first movers in the class.

Interestingly, Hewlett-Packard (HP), which has by far the largest market share in

this industry, does not appear to follow the same entry patterns as other dominant firms.

The second column of Table 3 has results from a specification that includes a dummy

16

variable for HP. The probability that HP introduces a product in a new submarket is

statistically indistinguishable from zero, and when it does enter, it has a much higher

probability of being an early mover.

It is also important to note that accounting for the selection effect of the entry

decision is quite important in understanding the patterns of dominant firm innovation in

this industry. To illustrate this concern which has been elucidated, but rarely addressed,

in the empirical literature, we present Models 3 and 4, which replicate the second stage

regressions from Models 1 and 2, but do not include the selection effects. Although the

coefficients on DOMINANT FIRM and HP are positive and statistically significant as

before, they are nearly twice the magnitude that they are in Models 1 and 2. To interpret

such a result as indicating that dominant firms are first movers into innovative segments

would be erroneous on two counts. First, as pointed out in Models 1 and 2, the dominant

firms are less likely to enter any given class of printers, thus the “first mover effect” is

true only for select classes of printers that dominant firms choose to enter. In general,

dominant firms are slower to enter innovative classes. Second, when they do enter,

unless one controls for the selection effect, the dominant firm first mover effect will be

substantially overstated.

Tables 4 and 5 present specifications to address our final question: how does entry

by a dominant firm affect competitors? We examine three effects: price, entry, and exit.

In the first column of Table 4, we regress the average wholesale price within a product

submarket one year after entry by a dominant firm, controlling for other market

conditions. Price falls substantially with competition from one or two dominant firms, by

far more than entry by a fringe firm. The coefficients on the dummy variables for the

17

number of dominant firms competing in the class are statistically significant at the 10%

level despite the small number of observations. In contrast, fringe firms would have to

introduce 30 competing products in the same niche to have the same effect on the average

price as the presence of one dominant firm. The second column of Table 4 regresses the

relative price of a new printer model to the average within a product class. Dominant

firms appear to price their products at approximately 14% over the average, and this

coefficient is significant at the 10% level. One interpretation of these results is that while

dominant firms can enjoy a price premium, their presence in a class heightens

competition among the fringe firms, which cut their prices in response to dominant firm

entry.

In Table 5, we present estimates from Poisson models of the count of new

products in a class, as well as the count of exiting products from a class, and include

dummy variables for whether one or two dominant firms were competing in the class one

year earlier. This allows us to examine the effect of dominant firm entry on market

structure and fringe firm behavior. Models 1a and 1b do not include demand growth in

the class as a control variable, while Models 2a and 2b do. The results indicate that the

presence of a dominant firm both increases entry and induces exit. It may be that

competitors are revamping their product lines in response to dominant entry, which

would cause them to withdraw current products and introduce new ones. It may also be

that entry by a dominant firm signals that the class is profitable, but fringe firms do not

account for entry by other fringe firms when making their own entry decision (or

systematically underestimate it). This would lead to “excess” entry, and thus also high

failure rates.

18

V. CONCLUSION

In the process of creative destruction, firms choose which markets to enter, and

when to enter those markets. These strategic choices are determined by what type of

innovation is being pursued (radical versus incremental); moreover, these strategic

choices can also have a substantial impact on the market structure in industries. In this

paper, we examine the desktop laser printer industry to determine how dominant firm and

fringe firm product launch strategies differ.

We show that dominant firms are quite selective in the submarkets they pursue.

In particular, they often stay out of the most innovative markets, and choose to enter

markets well behind the innovative frontier set by fringe firms. In our study, this is

especially true in the DPI dimension of printers. When dominant firms do enter markets,

they are usually early entrants, launching products before many fringe firms.

What is more interesting, though, is to what happens to markets after dominant

firm entry. Many antitrust scholars and policymakers are wary of dominant firm entry

into technology markets, fearing that such dominant entrants will wipe out fringe

competitors, leading to high market power for the entrant. However, the entry of the

dominant firm is followed by a wave of entry by other fringe firms. Thus, although exit

rates increase, this increase in exit is offset by an infusion of entrants that come in after

the dominant firm. The average price within a submarket declines more after entry by a

dominant firm than after entry by fringe firms. In addition, after entry by the dominant

19

firm, the demand in the submarket increases substantially.7 Nevertheless, the increase in

net entry and lower average prices suggest that entry by dominant firms does not harm

competition in this industry, at least in the short run. Consumers potentially benefit from

the presence of both fringe firms, which introduce more technologically advanced

products, as well as dominant firms, whose competitive effect appears to lower prices.

Consumers potentially benefit from the presence of both fringe firms, which introduce

more technologically advanced products, as well as dominant firms, whose competitive

effect appears to lower prices.

This paper contributes to our understanding of the behavior of dominant and

fringe firms in technologically advancing markets. In particular, we demonstrate that

dominant and fringe firms focus on different product segments and have different effects

on competition. The differences in the launch strategies of these types of firms have

important implications for the evolution of technology and market structure within an

industry.

7 Whether this is because dominant firms can better predict submarket growth, or because dominant firms cause submarket growth, is still an open question.

20

REFERENCES

Adner, R. and D. Levinthal (2001). “Demand Heterogeneity and Technology Evolution,” Management Science 47(5), 611-628. Bayus, Barry L., and William P. Putsis (1999). “Product Proliferation: An Empirical Analysis of Product Line Determinants and Market Outcomes,” Marketing Science 18(2): 137-153. Bond, R. S. and D. F. Lean (1977). “Sales, Promotion, and production Differentiation in Two Prescription Drug Markets,” Washington, D.C. U.S. Federal Trade Commission. Bowman, E. (1982), “Risk Seeking by Troubled Firms,” Sloan Management Review 23: 33-42. Bowman, Douglas, and Hubert Gatignon (1996). “Order of Entry as a Moderator of the Effect of the marketing Mix on Market Share,” Marketing Science 15(3): 222-242. Brittain, J. and J. Freeman (1980), “Organization Proliferation and Density-dependent Selection,” in J. Kimberly and R. Miles, The Organizational Life Cycle, San Francisco: Jossey-Bass, 291-338. Christensen, Clayton M. (1997). The Innovator’s Dilemma. (Boston: Harvard Business School Press).

Gilbert and Newbery (1982), “Preemptive Patenting and the Persistence of Monopoly,” American Economic Review, 72(3), pp. 314-26.

Glazer, A. (1985). “The Advantages of Being First,” American Economic Review 75: 473-80. Golder, Peter N. and Gerard j. Tellis (1993). “Pioneer Advantage: Marketing Logic or Marketing Legend?” Journal of Marketing Research 30: 158-70. Grabowski, Henry (1968), “The Determinants of Industrial Research and Development: A Study of the Chemical, Drug, and Petroleum Industries,” The Journal of Political Economy 76(2), 292-306. Hannan, M. and J. Freeman (1977), “The Population Ecology of Organizations,” American Journal of Sociology 82, 149-164.

Henderson, Rebecca (1993), “Underinvestment and Incompetence as Responses to Radical Innovation: Evidence from the Photolithographic Alignment Equipment Industry,” RAND Journal of Economics 24(2), pp. 248-270.

21

Henderson, R. and K. Clark, “Architectural Innovation: The Reconfiguration of Existing Product Technologies and the Failure of Established Firms,” Administrative Science Quarterly, Vol. 35, No. 1, Special Issue: Technology, Organizations, and Innovation. (Mar., 1990), pp. 9-30. Kamien, M. and N. Schwartz, “Timing of Innovations Under Rivalry,” Econometrica, Vol. 40, No. 1. (Jan., 1972), pp. 43-60. Kalyanaram Gurumurthy, William T. Robinson, and Glen L. Urban (1995). “Order of Market Entry: Established Empirical Generalizations, Emerging Empirical Generalizations, and Future Research,” Marketing Science 14(3): G212-G221. Katz, M. and C. Shapiro (1987), “R&D Rivalry with Licensing or Imitation,” The American Economic Review, 77(3), 402-420. Lambkin, M. (1988), “Order of Entry and Performance in New Markets,” Strategic Management Journal, Vol. 9, Special Issue: Strategy Content Research., 127-140. Lieberman, M. (1987), “Strategies for Capacity Expansion,” Sloan Management Review 28(4). Lieberman, Marvin B., and David B. Montgomery (1988). “First Mover Advantages,” Strategic Management Journal 9: 41-58.

Lerner, Josh (1997), “An Empirical Exploration of a Technology Race,” RAND Journal of Economics 28(2), pp. 228-247.

Makadok, Richard (1998). “Can First-Mover and Early-Mover Advantges be Sustained in an Industry with Low Barriers to Entry/Imitation?” Strategic Management Journal 19: 683-696. Mansfield, E., M. Schwartz, and S. Wagner (1981). “Imitation Costs and patents: An Empirical Study,” Economic Journal 91: 907-918. March, J. (1988), “Variable Risk Preferences and Adaptive Aspirations,” Journal of Economic Behavior and Organization 9: 5-24. Mitchell, W. (1989), “Whether and When? Probability and Timing of Incumbents' Entry into Emerging Industrial Subfields,” Administrative Science Quarterly, 34(2): 208-230. Mitchell, Will (1991). “Dual Clocks: Entry Order Influences on Incumbent and Newcomer Market Share and Survival When Specialized Assets Retain Their Value,” Strategic Management Journal 12(2): 85-100.

22

Moore, Michael J., William Boulding, and Ronald C. Goodstein (1991). “Pioneering Market Share: Is Entry Time Endogenous and Does it Matter?” Journal of Marketing Research 28: 97-104. Narasimhan, Chakravarthi, and V.John Zhang (2000). “Market Entry Strategy Under Firm Heterogeneity with Asymmetric Payoffs,” Marketing Science 19(4): 313-327. Nelson, R. and S. Winter (1982), An Evolutionary Theory of Economic Change, Cambridge, MA: Harvard University Press. Randall, Taylor, Karl Ulrich, and David Reibstein (1998). “Brand Equity and Vertical Product Line Extent,” Marketing Science 17(4): 356-379.

Reinganum, J. F. (1983), “Uncertain Innovation and the Persistence of Monopoly,” American Economic Review, Vol. 73, pp. 741-48.

Robinson, William T. (1988). “Source of Market Pioneer Advantages: The Case of Industrial Goods Industries,” Journal of Marketing Research 25: 87-94. Robinson, William T., Claes Fornell, and Mary Sullivan (1992). “Are Market Pioneers Intrinsically Stronger than Later Entrants,” Strategic Management Journal 13: 609-624. Schere, F. M. (1965), “Firm Size, Market Structure, Opportunity, and the Output of Patented Inventions,” American Economic Review, 55(5): 1097-1125 Schmalensee, Richard (1978). “Entry Deterrence in the Ready to Eat Breakfast Cereal Industry,” Bell Journal of Economics 9: 305-327. Singh, J. (1986), “Performance, Slack and Risk-taking in Organizational Decision Making,” Academy of Management Journal 29: 562-585. Tushman, M. and P. Anderson (1986), “Technological Discontinuities and Organizational Environments,” Administrative Science Quarterly 31: 439-465. Van De Ven, W. P. M. M., and B. M. S. Van Praag. "The Demand for Deductibles in Private Health Insurance." Journal of Econometrics, 17 (Nov. 1981), 229-252. Wernerfelt, B. and A. Karnini (1987), “Competitive Strategy Under Uncertainty,” Strategic Management Journal 8: 187-194.

23

Table 1: Variable definitions and summary statistics

Variable Obs Mean Std. Dev. Min Max Dummy variable for editorial award 3214 0.2526 0.4346 0 1 Units shipped 1641 12244.2300 48566.6700 0 650000 PPM (pages per minute) 3214 10.1696 4.2623 2 18 DPI (dots per inch) 3214 449.9517 260.8171 240 1800 PPM class 3214 3.8370 1.7159 1 6 DPI class 3214 1.5087 0.8292 1 4 Total # products in same DPI class 3214 193.5411 101.1172 1 320 Total # products in same PPM class 3214 107.9319 66.9080 1 255 Total # products in same DPI and PPM class 3214 43.3983 35.4591 1 127 Demand growth in class 3010 14.9657 167.4814 -1 2174.721 # firm's own products in same PPM class 3214 5.0669 4.3579 1 24 # firm's own products in same DPI class 3214 6.4406 5.4498 1 30 # firm's own products in same PPM-DPI class 3214 2.9502 2.1382 1 13 Total # models on market 3214 588.4661 185.6672 1 855 # new products in year 3214 168.4309 94.3193 1 416 List price of printer 2875 4018.0430 2921.1790 219 24950 Average list price within PPM-DPI class 2848 3983.1870 2107.0310 329 19950 Relative list price of printer to average in PPM-DPI class 2848 1.0216 0.6419 0.0931 13.5007 Market share 2910 0.0405 0.1111 0 0.6351 Dominant firm dummy (at least 10% share) 3379 0.0769 0.2665 0 1 Number of dominant firms in PPM-DPI class 3192 0.6764 0.6733 0 2 # new products in PPM-DPI class 3571 6.3957 6.6100 0 26 # exiting products from PPM-DPI class 3571 9.4060 11.9100 0 46

24

Table 2: Technological position Panel 1

DPI PPM

Best products while dominant Mean Std. Dev. Mean Std. Dev. APPLE_COMPUTER_CO 300 0 8 0 FRINGE FIRMS 1370.88 355.15 17.94 0.66 FUJITSU_AMERICA_INC 600 0 18 0 HEWLETT-PACKARD_COMPANY 547.83 114.97 16.02 2.83 IBM/LEXMARK 1200 0 16 0 Best products over all years for firms dominant in any year APPLE_COMPUTER_CO 533.33 126.49 14.22 4.02 FRINGE FIRMS 1368.06 354.87 17.94 0.63 FUJITSU_AMERICA_INC 648.39 233.63 18.00 0.00 HEWLETT-PACKARD_COMPANY 542.55 119.32 15.85 3.04 IBM/LEXMARK 924.59 356.21 14.82 1.84 NEC_TECHNOLOGIES_INC 528.00 130.77 9.28 0.98 Panel 2 Regression of Y = Best DPI/PPM of firm's portfolio DPI PPM

Coef.

(Std. Err.) Coef.

(Std. Err.) Dominant firm -827.119** -4.983** (88.715) (1.657) Total # models in mkt 1.374** 0.005 (0.353) (0.007) Total # new models on mkt -0.242 -0.012 (0.697) (0.013) Constant 500.329* 18.370**

(234.968) (4.388) N 25 25 Adj Rsq 0.845 0.2981 FE Year Year Panel 3 Years at

DPI frontier

Years at PPM

frontier LASERMASTER 7 KYOCERA_UNISON 11 PRINTWARE 7 FUJITSU 10 XANTE 5 LEXI_CORP 9 DEC 3 JRL_SYSTEMS 6 GCC_TECHNOLOGIES 3 OLYMPUS 5 HEWLETT-PACKARD 3 PRINTER_SYSTEMS 5 NEWGEN_SYSTEMS 3 HEWLETT-PACKARD 3 QMS_INC 3 SAMSUNG 3

25

Table 3: Order of entry Model 1 Model 2 Model 3 Model 4 Second Stage: Y = 1 if firm is the first mover in class

Coef. (Std. Err.)

Coef. (Std. Err.)

Coef. (Std. Err.)

Coef. (Std. Err.)

Dominant firm 0.480* 0.821** (0.217) (0.110) HP 0.715* 1.157** (0.282) (0.132) Engine manufacturer dummy -0.187 -0.166 -0.238** -0.172* (0.121) (0.127) (0.075) (0.076) # firm's models in same ppm class 0.073** 0.077** 0.068** 0.074** (0.020) (0.021) (0.009) (0.009) # firm's models in same dpi class -0.028 -0.031 -0.035** -0.037** (0.015) (0.016) (0.008) (0.008) Constant 0.847** 0.765* -0.572** -0.630** (0.267) (0.305) (0.092) (0.094) FE Class Class Class Class First Stage: Y = 1 if firm enters class

Dominant firm -0.278* (0.120) HP -0.199 (0.148) Editorial award for one of firm's products 0.362** 0.357**

(0.068) (0.072) Rebrand dummy 0.088 0.101 (0.096) (0.098) Engine manufacturer dummy 0.078 0.082 (0.066) (0.068) Total number of models in market -0.001** -0.001** (0.000) (0.000) # firm's models in same dpi class -0.005 -0.006 (0.007) (0.007) # firm's models in same ppm class -0.013 -0.015 (0.009) (0.008) Constant -0.576** -0.595** (0.128) (0.128) /athrho -1.210** -1.117** (0.319) (0.333) N 3032 3032 3021 3021 Log L -1579.69 -1579.27 -1116.23 -1105.05 FE Class Class

26

Table 4: Effect of dominant entry on price Average price Relative price

Variable Coef.

(Std. Err.) Coef.

(Std. Err.) Dominant firm 0.1483 (0.0823) One dominant firm in class -854.23 (501.85) Two dominant firms in class -1,611.61 (885.17) Total # models in mkt -4.73** 0.0002 (1.77) (0.0002) Total # competing models in same ppm class 25.17** -0.0022 (5.70) (0.0042) Total # competing models in same dpi class 2.42 -0.0003 (3.27) (0.0007) Total # competing models in same ppm-dpi class -26.70 0.0001 (15.39) (0.0004) Demand growth in class -0.54 0.0003 (0.63) (0.0013) Constant 4,934.96** 0.8927** (877.04) (0.1462) N 52 1154 Adj Rsq 0.3671 0.0217

FE Year

Year Model's order of

entry

27

Table 5: Effect of dominant entry on class entry and exit Model 1 Model 2

Poisson of Y = # entering/exiting products in class Entry Exit Entry Exit

Variable Coef.

(Std. Err.) Coef.

(Std. Err.) Coef.

(Std. Err.) Coef.

(Std. Err.) One dominant firm in class 0.513** 0.353** 0.333** 0.144 (0.109) (0.106) (0.115) (0.111) Two dominant firms in class 0.578** 0.451** 0.458** 0.228* (0.129) (0.114) (0.141) (0.119) Total # models in mkt -0.002** 0.012 -0.003** 0.004** (0.001) (0.237) (0.000) (0.001) Total # new models on mkt 0.006** 0.009 0.004** -0.004** (0.001) (0.382) (0.001) (0.001) Total # competing models in same ppm class 0.003** 0.001 0.001 0.001 (0.001) (0.001) (0.001) (0.001) Total # competing models in same dpi class 0.001** 0.005** 0.000 0.005** (0.001) (0.001) (0.001) (0.001) Total # competing models in same ppm-dpi class 0.006** 0.016** 0.008** 0.015** (0.002) (0.002) (0.002) (0.002) Demand growth in class 0.000** 0.000 (0.000) (0.000) Constant -0.013 -11.450 1.524** -0.922* (1.002) (300.639) (0.223) (0.522) N 176 176 126 126 Log L -439.752 -422.606 -336.753 -334.971

Figure 1: Number of Firms and Products in Marketplace

Num

ber o

f Mod

els

Number of Products and FirmsYear

Num

ber o

f Firm

s

Models Firms

1984 1986 1988 1990 1992 1994 1996

1

633

1

230

Figure 2: Product Entry and Exit

Num

ber o

f New

Mod

els

Entry and Exit, 1984-1996yyyy

Num

ber o

f Exi

ting

Mod

els

Entry Exit

1984 1986 1988 1990 1992 1994 1996

0

20

40

60

80

100

120

140

160

180

200

0

20

40

60

80

100

120

140

160

180

200

Figure 3: Product Distribution and Classes

dpi

ppm0 5 10 15

0

500

1000

1500

2000

Class 1

Class 2

Class 3

Class 4

Class 5

Class 6

Class 7

Class 8

Class 9

Class 10

Class 11

Class 12

Class 14

Class 14

Class 15

Class 16

Class 17

Class 18

Class 19

Class 20

Note: Each small circle represents a printer.

Figure 4: Speed of printers by Firm Type by Year

yyyy

ppmfringe ppmdom

1988 1997

2

18

Figure 5: Resolution of printers by Firm Type by Year

yyyy

dpifringe dpidom

1988 1997

240

1800